Dynamic Eyebox Steering for Improved Pinlight AR Near-Eye Displays
View abstract on PubMed
Summary
This summary is machine-generated.This study enhances augmented reality (AR) near-eye displays (NEDs) by using 3D pupil tracking to enlarge the eyebox, improving user experience for AR systems.
Area Of Science
- Optics and Photonics
- Human-Computer Interaction
- Augmented Reality Displays
Background
- Optical-see-through near-eye displays (NEDs) for augmented reality (AR) present trade-offs in field of view (FOV), eyebox size, and form factor.
- Existing AR NED technologies struggle to balance key performance metrics for optimal user perception.
- Compact, wide-FOV pinlight AR NEDs are limited by small, unusable eyeboxes without advanced compensation techniques.
Purpose Of The Study
- To enhance a compact, wide-FOV pinlight AR NED design by improving eyebox usability.
- To implement real-time 3D pupil localization for dynamic eyebox steering.
- To validate the effectiveness of the proposed eyebox steering method through user studies.
Main Methods
- Developed a dual-camera rig for stereoscopic pupil capture and real-time 3D pupil localization.
- Integrated 3D pupil data to dynamically calculate and modulate display patterns for spatio-temporal light steering.
- Designed and conducted a user study to quantify the improvement in eyebox usability with the implemented steering method.
Main Results
- The eyebox steering method significantly reduced instances where users could not perceive the test pattern (from 10.5% to 1.25%).
- A demonstrable compact prototype incorporating the dynamic eyebox steering was successfully built.
- User study results confirmed the practical effectiveness of the real-time 3D pupil localization and eyebox steering technique.
Conclusions
- Dynamic eyebox steering via real-time 3D pupil localization is a crucial advancement for practical AR NED usability.
- This method enhances simple wide-FOV pinlight NEDs, moving them beyond demonstration prototypes.
- Detailed display design, calibration, and user study methodologies are provided to benefit future NED research.

